Skip to main contentAbout StateSet Sandboxes
StateSet Sandboxes provide isolated execution environments for AI agents with controlled access to tools and data.
How sandboxes work
- A controller provisions a sandbox on demand.
- Your agent executes commands and file operations in an isolated runtime.
- Outputs stream back to your client for auditing and traceability.
Why this design
Isolation reduces risk while enabling agent workflows to run in production environments.
When to use sandboxes
- Agent evaluation: test workflows safely before production
- Production runs: execute with clear audit trails
- Multi-tenant setups: isolate workloads by org
Relationship to other features
Sandboxes power agent execution for ResponseCX, Console, and iCommerce workflows.
Further reading
